Pesentation slides

1. Wednesday, 12 October 2016
Morphological model of the River Rhine
branches in The Netherlands
Mohamed F.M. Yossef
mohamed.yossef@deltares.nl

2. 15 October 2016
Motivation
Vuren Nijmegen Erlecom
Hulhuizen
St.-Andries
Gorinchem
Dodewaard
Ochten
Tiel
• The Rhine River is considered the backbone of
the Northwest European waterways network.
• Efforts are made to maintain and improve the
navigation channel.
• A need for a tool that enables:
• analysis of historical trends,
• prediction of future trends,
• evaluating different measures
 a numerical model,
that is accurate and (fast)

3. 12-01-2011
DVR
• Client: Rijkswaterstaat Oost-Nederland: the river manager
• Target: maintain a sustainable navigation channel NL+D.
• The project is called “Duurzame Vaardiepte Rijndelta”
(Sustainable Navigation Channel Rhine delta)

4. Outline
• Grids
• Schematization
• Time management
• Dredging
Construction
• Hydrodynamics
• Morphology
• Dredging
Calibration
• Effect of longitudinal dams
• Dredging the navigation channel
• Sediment nourishment
experiment
Application
15 October 2016

5. 15 October 2016
Project phases
• Phase 1 (2005)
• Model construction
• Development, implementation and testing of innovative aspects
• Phase 2 (2006)
• Primary calibration
• Case studies
• Phase 3 (2007)
• Model operationalization: Calibration & optimise for speed
• Phase 4 (2008)
• The operational model  further improvements
• Testing measures (by different consultants)
• 2009
• Extension to complete all branches
• Testing measures (different consultants, supported by Deltares)
• 2010
• update of the model (extend the hydrograph)
• Testing several measures (different consultants , supported by Deltares)
• 2011
• Testing measures (different consultants , supported by Deltares)
• 2012-2014
• Improvements and Application to give advice (Deltares & different consultants)
• 2015
• Extension to graded sediment and nourishment testing (Deltares)
• 2016-2017
• Migration to flexible mesh
Initial phase

6. Model Construction

7. 15 October 2016
Model construction  Extent
The model is around 360 km long, covering:
Bovenrijn
km 853-867 (Emmerich – Pannerdensche Kop)
Waal
km 867-953 (Pannerdensche Kop –Werkendam)
Pannerdensche Kanaal
km 876-879 (all Pannerdensche Kanaal)
IJssel
km 879-912 (IJsselkop – Doesburg);
Nederrijn
km 879-889 (IJsselkop – Driel).
(1)
(2)
(3)
(4)

8. Ijssel Kop
Merwedes
Pannerdensche Kop
Ketelmeer  IJssel
15 October 2016
Model construction  Domains
Use of domain decomposition at bifurcations

9. 15 October 2016
Pannerdensche Kop
Bovenrijn
Waal
IJsselkop
Nederrijn
8 – grid cells
IJssel
8 – grid cells
Waal – a
Waal – b
10 - grid cells
12 - grid cells
8 - grid cells

10. Example grids  impact on results

11. Example grids  impact on results

12. Example grids  impact on results

13. Example grids  impact on results

14. Example grids  impact on results

15. 15 October 2016
Model construction  Grid characteristics
Gridname Bovenrijn Waal – part a Waal – part b Waal – part c Pannerdensch
Kannal
number of grid cells
≈ 68 000
55x177
9735
47x296
13616
47x401
18847
47x353
16591
67x137
9179
main channel
number of grid cells 10 12 12 12 8
grid cell width (m) ≈ 34 23 21 26 20
grid cell length (m) ≈ 80 80 80 80 80
aspect ratio 1:2.4 1:3.4 1:3.8 1:3 1:4

16. 15 October 2016
Model construction  Schematisation
• GIS stored information (BASELINE: ArcView add-on)
• Reference schematisation including:
• bed topography,
• hydraulic roughness,
• Structures:
> barriers
> groynes (as weirs),
> dikes
> steep obstacles in the floodplains.
• Initial topography (year 1997), replaced by the multi-beam measurements
(1999 for the Waal and Bovenrijn, 2002 for the Pannerdensche Kanaal)
• Non-erodible layers added

17. 15 October 2016
• Main channel:
• Roughness is based on alluvial roughness (dune height
prediction)
• Roughness specified per river reach (between measurement
stations) and variable in longitudinal direction (based on
hydraulic calibration)
• Avoid sharp transitions in roughness
• Floodplain
• Roughness based on vegetation coverage, with analytical
model to calculate vegetation roughness.
Model construction  Roughness

18. 15 October 2016
Model construction  Structures
• Groynes, weirs, other obstacles: all defined in the model as weirs.
• Measures such as groyne removal, lowering, shortening or
extending can be simulated easily.
Initial bed level, weir heights, and flow pattern showing depth-averaged
velocity

19. Time management
Discharge schematization
Simulation management tool

20. 15-Oct-16 22
Discharge schematisation  upstream boundary
0
2000
4000
6000
8000
10000
12000
14000
16000
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035
Date
Q
(m3/s)
4
1
2
3
4

21. 15-Oct-16 23
0
1000
2000
3000
4000
5000
6000
7000
8000
1993 1994 1995 1996
Time
Q
(m
3
/s)
0
1000
2000
3000
4000
5000
6000
7000
8000
1993 1994 1995 1996
Time
Q
(m
3
/s)
Q1 Q3
Q2
0
1000
2000
3000
4000
5000
6000
7000
8000
1993 1994 1995 1996
Time
Q
(m
3
/
s
)
/morfac
• Repeat a yearly schematised hydrograph using a sequence of steady
discharges
• Apply a “morphological acceleration factor” to speed up morphology
(same morphological changes in shorter flow period): factor 50 – 200
• morfac is inversely proportional to discharge
Running long simulations: quasi-steady & accelerated
Morphological time Hydrodynamic time

22. 15-Oct-16 24
Q1
restart
file
Q2
database
restart
file
Q3
restart
file
hydrodynamic parameters
morphodynamic parameters
Operated using Python scripts
ETCETERA
Simulation management tool  running long simulations

23. Dredging

24. 15 October 2016
Dredging and dumping module
Functionality
• dredging within user defined polygons (arbitrary number)
• may encompass large or small areas (flexible)
• dumping within user defined polygons (arbitrary number)
• distribution of dredged material over dumping areas given by user in
advance (new)
• dredging and nourishment intervals (new)
• sequential dumping in a series of dumping blocks (new)
30%
70%

25. 15 October 2016
Dredging and dumping  criterion
Trigger dredging when:
• Bed level above threshold
a) threshold level prescribed
(spatially varying threshold
level allowed), or
b) constant (per polygon) water
depth below specified
reference water level
(spatially varying water level
allowed e.g. OLR)
• with, a given dredging rate
• externally provided sediment
(nourishment)
(a)
clearance
threshold level
(b)
reference level
clearance
threshold level
constant per
dredge area

26. 15 October 2016
(c)
(b)
Dredging and dumping  method/options
Options
• dredge method
a) dredge top first,
b) dredge proportional,
c) dredge uniform
• dumping method
a) dump deepest first,
b) dump uniform
• dredge time constraints
• only within certain period
• minimum time since previous
dredge action
• only below a certain minimum
water depth
(a)
(a) (b)

27. Calibration
• Hydrodynamic calibration
• Choose a transport formula suitable for all branches.
• morphological calibration of the model
• Calibration for the dredging activities in the Waal.

28. 15 October 2016
Hydraulic calibration
- Discharge distribution
- Water levels
Tips:
1st step  split branches;
use desired discharge per branch
start from the downstream end
avoid large jumps in roughness values

29. 15 October 2016
Transport formula
Requirements
• The formula should have a similar behaviour as the MPM formula for
Shields parameter values below 0.09, which corresponds to the
conditions in the Bovenrijn.
• The formula should have a similar behaviour as the EH formula for
Shields parameter values above 0.3, which corresponds to the conditions
in the Midden-Waal and the Beneden-Waal.
• For physical reasons, n in (S = mun) should always be larger than 3.
Preferably, the degree of nonlinearity should decrease monotonously as
the Shields parameter increases.
• n should be about 4 or 5 for large Shields parameter values. The value of
5 complies with the EH predictor.

30. 15 October 2016
Transport formula
 
3 2
3
50
8
MPM MPM MPM cr
MPM
S m
m g D
  
 
 
5
3 2
50
0.05
EH EH EH
EH
S m u
m
g C D




P
cr
P



 
  
 
       
   
1 5 2
3 2 1 1 5 2
2
50
P
P P P P
a
AS MPM MPM cr EH EH
MPM EH
S m m C D

   
    
 
   
    
 
1
b
AS P MPM P EH
MPM EH
S S S
 
    
Engelund & Hansen, (1967)
preferred for lower parts
Meyer-Peter & Müller, (1948)
preferred for upper parts
Combined formula  test
van Rijn (1984)
VR BED BED SUS SUS
S S S
 
   
Tested Formulations
A Van Rijn (1984)
The formula of Van Rijn (1984) takes the form:
s b
S = +
S S (A.1)
where:
3 0.3 2.1
50 *
3 0.3 1.5
50 *
0.053 for 3.0
0.1 for 3.0
b
g D D T T
S
g D D T T


  

 
 


(A.2)
First the bed-load transport expression will be explained. In Eq. A.2 T is a dimensionless bed shear parameter, written as:
bc bcr
c
bcr
-
T =
  

(A.3)
It is normalised with the critical bed shear stress according to Shields (bcr), the term c bc
  is the effective shear stress. The formulas of the shear stresses are:
1
8
2
bc w cb
= f u

 (A.4)
 
2
10
0.24
log 12 /
cb
c
=
f
h 
 
 
(A.5)
 
2
10
18log 12 / c
c
h
=
C


 
 
 

 
(A.6)
where Cg,90 is the grain related Chézy coefficient:
10
90
12
18log
3
h
C
D
 
   
 
(A.7)
The critical shear stress is written according to Shields:
50
bcr cr
w
= g D
 
  (A.8)
in which cr is the critical Shields parameter for initiation of motion, which is a function of the dimensionless particle parameter D:
1
3
* 50 2
g
D D


 
  
 
(A.9)
The suspended transport formulation reads:
s a
cs
= u h
f
S C (A.10)
In which Ca is the reference concentration, u depth averaged velocity, h the water depth and fcs is a shape factor of which only an approximate solution exists:
0
1
if 1.2
( ) if 1.2
c c
cs
c c
( ) z
f z
f =
f z z





(A.11)
 
   
   
1.2
0
/ /
1 / 1.2
c
c
z
c c
c z
c c
h h
f z
h z
 



 
(A.12)
   
1.2
1
/
log /
1 /
c
c e c
c
h
f z h
h



 
  

 
(A.13)
where c is the reference level or roughness height (can be interpreted as the bed-load layer thickness) and zc the suspension number:
*
min 20, s
c
w
=
z
u

 
 

 
 
(A.14)
*
8
cb
f
u =u (A.15)
2
*
min 1.5,1 2 s
w
=
u

 
 
 
  
 
 
 
(A.16)
0 8 0 4
*
2 5
0 65
. .
s a
w C
= .
u .

   
   
 
 
(A.17)
The reference concentration is written as:
1.5
50
1 0.3
*
0.015
a
c
d T
C
D


 (A.18)
The following formula specific parameters have to be specified as input to the model.

31. 15 October 2016
Transport formula – offline analysis
note:
αEH = 0.5;
reduced van Rijn equation with
ws = f(d50), αSUS = 0.3, αBED = 1.5.
transport rate
dimensionless transport rate
degree of nonlinearity
the formula of van Rijn yields the
desired behaviour

32. 15 October 2016
Transport formula
• We favour using the formula of van Rijn (1984).
• The implementation in Delft3D is modified such that:
• it is possible to calibrate bed load and suspended load
separately (αBED and αSUS),
• it is possible to opt for a reduced formula; total load: suspended
load is added to bed load (no advection-diffusion equation for
suspended load)  using option ‘bedload’
• it is possible to use a constant critical Shields parameter for the
initiation of motion; user defined as a calibration parameter.
Conclusions:

33. 15-Oct-16 35
Calibration of the morphodynamic model
1D-behaviour:
• annual sediment transport
volumes/rates,
• bedforms celerity (highest priority)
• annual bed level changes, and
• period-averaged bed level
gradient.
Parameters:
•  (coefficient in transport formula)
• cr
• D50 = f(x)
• D90 ≈ 4 × D50
Procedures:
• offline calculations (100s)
• online calculations (10s)
2D-behaviour (bar-pool patterns):
• transverse slopes in bends, and
• position of crossing between two
opposite bends
Parameters:
• coefficient affecting the spiral flow
intensity due to curvature (Espir)
• coefficient influencing the effect of
transverse bed slope (Ashld).
Procedures:
• online calculations (10s)
we are using the sediment transport formula of van Rijn (1984), with a tweak.

34. 15 October 2016
Morphological boundary conditions
• The following morphological boundary conditions are available:
1. free bed level, i.e. bed level change at boundary equals the
internal bed level change (not recommended).
2. fixed bed level, (default)
3. prescribed bed level variation [m]
4. prescribed bed level change rate [m/s]
5. prescribed sediment transport rate with pores (sand volume)
[m3/s/m], and
6. prescribed sediment transport rate without pores (stone
volume) [m3/s/m]

35. 15 October 2016
Calibration result – 1D behaviour
Bed-form celerity
cross-section and reach-averaged yearly bed level change cross-section and reach-averaged bed level gradient
Bovenrijn
Boven
Waal
Midden
Waal
Benden
Waal
Pannerdensch
Kanaal
Bovenrijn Boven
Waal
Midden
Waal
Benden
Waal
Pannerdensch
Kanaal
Firstly, sediment budget for the Rhine branches

36. Calibration result – 1D (trench)
15 October 2016
Bed celerity ca. 1 km/year  using 3 trenches

37. 15 October 2016
1 km/year
Calibration result – 1D (trench)

38. 2D calibration
15 October 2016
longitudinal profiles along left and right banks

39. Dune height – for dredging
15 October 2016
Settings
Bdf = Y
BdfMor = Y
BdfH = FredsoeMPM
BdfL = vanRijn84
BdfR = vanRijn84
BdfEps = 0.8
BdfRlx = THConst
BdfT_H = 57600
BdfADV = N
BdfThetaC = 0.047
• Four dune height predictors have
been implemented: Van Rijn
(1984c), Fredsøe (1982) or
Engelund and Hansen (1967), and a
power relation.
• Temporal and spatial variations;
implemented by means of an
advection relaxation equation.
 results

40. 12 maart 2009
• Dredging and dumping according to prescribe
criteria
• Dredging in specified areas and dumping at a
prescribed distance upstream
PolygonFile = Dredge150b.pol
MaxVolRate = 1.0E16
DredgeDepth = 2.80
Clearance = 0.30
MinimumDumpDepth = 2.5
AlphaDuneHeight = 0.5
DredgeDistr = 2
DumpDistr = 2
Inpolygon = 1
Dredging model

41. 12 maart 2009
Dredging model
• Comparison with data 2000-2002: confirms that Upper Waal
dominated by dredging in bends (structural), and the Middle Waal
due to dunes (incidental)
structural dredging:
• Hulhuizen (km 870)
• Erlecom (km 875)
• Haalderen (km 880)
• Nijmegen (km 885)
• St. Andries (km 928)
incidental dredging:
in the Midden Waal
between km 887-915

42. 15 October 2016
Rhine Model – overall results
2D-behaviour (40-years)
2D-behaviour (detail)
1D-behaviour (40-years)
Dune heights

43. Rhine Model – 1D behaviour
15 October 2016
1D-behaviour (40-years)
Cross-section averaged bed level
black  initial

44. Rhine Model – 2D large-scale
15 October 2016
2D-behaviour (40-years)
deposition
erosion

45. Rhine Model – 2D local
15 October 2016
2D-behaviour (detail)
depth (m)
Black lines define navigation channel
red colors too shallow

46. Case studies
• Effect of longitudinal dams
• Dredging the navigation channel
• Sediment nourishment experiment

47. 15 October 2016
Effect of longitudinal dams
Function:
•constriction of main channel during
discharges below bank full
•increase flood conveyance capacity
during floods
Parallelwerk Walsum-Stap, Rijn km 793,5 – 795.
Wasser- und Schiffahrtsverwaltung des Bundes,
Wasser- und Schifffahrtsdirection West
Longitudinal dams to replace groyne lowering

48. 15 October 2016
Effect of longitudinal dams
The effect is due to:
• channel constriction,
• discharge extraction and supply.
Main conclusion:
• The local effect of longitudinal
dams are rather significant.
 This calls for optimization of
the inflow and outflow sections,
 we need an additional
analysis tool
erosion
deposition
erosion
deposition

49. 15 October 2016
Dredging the navigation channel
Evaluation of dredging volumes for:
• Navigation channel of 150 m x 2.50 m (Case B150)
• Navigation channel of 170 m x 2.50 m (Case B170)
• Navigation channel of 150 m x 2.80 m (Case B152)
This called for additional functionalities to be implemented:
• Dredging and Nourishment intervals
• Sequential dumping in a series of dumping blocks
• Dredging considering dune heights (not utilised yet)

50. 15 October 2016
Dredging the Navigation Channel
Dredging from the navigation channel
and dumping in the normal line; with a
1.0 km unit.
dredging

51. 15 October 2016
Morphological response to dredging
simulation of 10 years (bed level difference)
see the effect of dredging near the end
red is dumping
blue is dredging
Dredging also affects the transverse cross-section, it is not a
simple 1D problem
cross profile

52. 15 October 2016
Effect of channel dimensions on dredging volumes
width = 150m & depth 2.50m:
43,000 m3
width = 170m & depth 2.50 m:
700,000 m3 (x16 times)
width 150m & depth 2.80 m:
290,000 m3 (x7 times)

53. 15 October 2016
Sediment nourishment
photos Rees
Unloading sand and gravel from split-barges
Key parameters for success are:
• Quantity
• Location and section-length
• Composition of sand/gravel
mixture
Modeling is needed for optimization!

54. 15 October 2016
Sediment nourishment
longitudinal profile  increases
cross profile  decreases
feeding with coarser material
Stop degradation by levelling-off
the sediment-transport gradient
Increase efficiency by dumping
coarse sediment (coarser than
original bed composition)

55. 15 October 2016
Sediment nourishment  test Bovenrijn
• amount = 6000 ton/week
• dump instantaneously
• total 150.000 m3
• thickness of layer about 0.3 m

56. Sediment nourishment behaviour
15 October 2016
• dump instantaneously
• total volume 150.000 m3
• thickness of layer about 0.3 m
z = “Bed-level Nourished (t)” minus “Bed-level Reference (t)”

57. Result - Nourishment test near Lobith
spreading of tracer fraction
V=150,000 m3

58. Result - Nourishment test near Lobith
spreading of tracer fraction

59. Tracer. 2
0.5 < D < 1 mm
(fine)
Tracer. 7
8 < D < 16 mm
(coarse)

60. 15 October 2016
Propagation speed of different fractions
Fine
Medium
Coarse
partial mobility

61. Yearly repeated feeding (150,000 m3/yr)

62. 15 October 2016
Overall conclusions
• We have a morphological model that covers the Rhine Branches in the
Netherlands.
• Well calibrated
• Hydrodynamics
• 1D morphological behaviour  bed celerity
• 2D morphological behaviour  bar-pool pattern
• Unique model:
• Large-scale, yet detailed
• Refined dredging and dumping
• Able to simulate different types of measures
• rather fast, 40 years in 4 days.
• The model can be used effectively for evaluation of the effect of different
engineering measures with continuous application in projects
• The model is disseminated to all consultants in the Netherlands for
application in different projects
• Processes extended to include graded sediment (not discussed today)

63. Moving to flexible mesh – starting
• More flexibility
• Same model for different types of studies (hydrodynamics,
morphology, water quality); mixed resolution
15 October 2016

64. 15 October 2016
Moving to flexible mesh – starting

65. 15 October 2016
Moving to flexible mesh – starting

66. 15 October 2016
color  vorticity

67. Delft Software Days 2016
 600+ participants
 200+ organisations
 50+ countries
 3 symposia &
user meetings
 3 workshops
 17 courses
Course
Delft3D Flexible Mesh – River modeling
26-27 October 2016
 1D and 2D river hydrodynamics
 Real-time control
 Automatic calibration with OpenDA
 Python scripting
www.dsd-int.nl